Everything AlphaFold tells us about protein knots

Area of Science:

• Structural Biology
• Computational Biology
• Bioinformatics

Background:

• Machine learning (ML) advances protein analysis, overcoming empirical research limitations.
• The AlphaFold Database provides extensive, high-quality protein structure predictions.

Purpose of the Study:

• To identify common trends and characteristics of knotted proteins.
• To investigate the distribution and prevalence of protein knots across different domains of life.

Main Methods:

• Analysis of all high-quality protein predictions from the AlphaFold Database.
• Identification and classification of potentially knotted protein models.
• Comparative analysis of knotted protein prevalence across Bacteria, Eukaryota, and Archaea.

Main Results:

• The majority of identified knotted proteins exhibit a 31 knot topology.
• The prevalence of knotted proteins is approximately 0.4% across all proteomes, irrespective of the domain.
• Organismal living conditions do not significantly affect the number of knotted proteins.
• Four protein families (SAM synthase, TrmD, TrmH, RsmE methyltransferases) are universally found in Bacteria.

Conclusions:

• Protein knotting is a widespread phenomenon, not specific to certain domains or environmental conditions.
• Specific protein families consistently display knotting, suggesting functional or evolutionary significance.